Compaction behaviour of soils

Abstract

Soil compaction is widely applied in geotechnical engineering practice. It is used to maximise the dry density of soils to reduce subsequent settlement under working loads or to reduce the permeability of soils. The durability and stability of structures are highly related to the appropriate compaction achievement. The structural failure of roads and airfields, and the damage caused by foundation settlement can often be traced back to the failure in achieving adequate compaction. For that reason, soil compaction is important for engineering activities involving earthworks. Compacted soils are unsaturated by nature, which includes both air and water within their voids. Thus, unsaturated soil mechanics principles are crucial in understanding the compaction behaviour of soils. There are several qualitative studies, which attempt to explain the compaction behaviour of soils and there is a vast body of literature covering the behaviour of compacted soils. Still, fundamental research on the compaction process is limited. In addition, the current constitutive models available for unsaturated soils assume that the soil state after compaction is the initial state of the soil. However, compacted soils undergo a stress history which influences the post compaction behaviour. Considering these facts, it still remains that the compaction of soil is a complex phenomenon, which is not well explained, particularly from a quantitative sense. Further understanding of the compaction behaviour during the compaction process will provide important insights on the behaviour of compacted soils. The main aim of this research project is to extend the current understanding of the compaction process of soils. The research focuses on three different areas: investigating the experimental behaviour of soils during the static compaction process and obtaining data for compaction modelling; developing a compaction model using the existing constitutive models for unsaturated soils; and evaluating the performance of this model in predicting the compaction behaviour of soils. In the experimental part, static compaction tests were conducted on two different granular soils, sand with 2% and 5% bentonite content by weight. The tests were undertaken on samples with different water contents in order to observe the effect of matric suction on the compaction behaviour. The initial matric suction of the specimens was measured using the null type axis translation technique and the matric suction variations were monitored during the compaction process. It was found that the unsaturated samples were always more compressible than the saturated sample. This finding is contrary to the assumption made in most constitutive models, and thus modelling the compaction behaviour using these models may result in some deficiencies. In addition, in granular soils with low water content the axis translation technique was found to be very time consuming for the suction measurements. This was attributed to the discontinuous water phase within the samples. To develop a compaction model, a volume change constitutive relationship for unsaturated soils, defined in terms of two independent stress variables, was incorporated with pore pressure predictions. The model was developed for undrained, semi-drained and drained loading conditions. Initially, compressibility coefficients in the volume change relationship were considered as constant parameters, i.e., the compressibility of a soil element does not change with increasing vertical stress. Using constant compressibility coefficients, the compaction curve can be predicted only for the wet side of the curve, not the dry side. Thus, variable compressibility coefficients were derived from constitutive models proposed in the literature, and using these coefficients, the well-known shape of the compaction curve was predicted on both dry and wet side of the compaction curve. It was found that the shape of the compaction curve can be the theoretically predicted using unsaturated soil mechanics principles. The main insight gained from the model development was that the influence of matric suction on the material compressibility with respect to net stress is the governing factor determining the inverted parabolic shape of the compaction curve. The performance of the compaction models were examined on their ability to predict the compaction behaviour of soils. Data for four different soils, two sand-bentonite mixtures tested in this study, and Boom clay and Speswhite kaolin data from literature, were used for models evaluation. Two different constitutive modelling approaches were analysed, which are the separate stress state variables approach and combined stress state approach. It was concluded that the samples prepared from initially slurry soils and from initially dry soils could not be treated the same and would require the use of different sets of soil parameters. In addition, the compaction behaviour of soils, prepared from initially dry samples, could only be modelled over a narrow range of water contents using a single set of soil parameters. Minimum two sets of soil parameters are required to model the compaction behaviour over a wide range of water contents with the current constitutive models.